Occurrence of Contaminant Accumulation in Lead ... - ACS Publications

Ohio 45268, and Pegasus Technical Services, Inc.,. 46 E. Hollister Street ... and Ra, have been shown to accumulate in drinking-water distribution sys...
33 downloads 3 Views 10MB Size
Environ. Sci. Technol. 2008, 42, 4285–4291

Occurrence of Contaminant Accumulation in Lead Pipe Scales from Domestic Drinking-Water Distribution Systems M I C H A E L R . S C H O C K , * ,† ROBERT N. HYLAND,‡ AND MEGHAN M. WELCH‡ United States Environmental Protection Agency, Office of Research and Development, National Risk Management Research Laboratory, Water Supply and Water Resources Division, 26 West Martin Luther King Drive, Cincinnati, Ohio 45268, and Pegasus Technical Services, Inc., 46 E. Hollister Street, Cincinnati, Ohio 45219

Received October 2, 2007. Revised manuscript received December 13, 2007. Accepted December 17, 2007.

Previously, contaminants, such as Al, As, and Ra, have been shown to accumulate in drinking-water distribution system solids. Accumulated contaminants could be periodically released back into the water supply causing elevated levels at consumers’ taps, going undetected by most current regulatory monitoring practices and consequently constituting a hidden risk. The objective of this study was to determine the occurrence of over 40 major scale constituents, regulated metals, and other potential metallic inorganic contaminants in drinking-water distribution system Pb (lead) or Pb-lined service lines. The primary method of analysis was inductively coupled plasma-atomic emission spectroscopy, following complete decomposition of scale material. Contaminants and scale constituents were categorized by their average concentrations, and many metals of potential health concern were found to occur at levels sufficient to result in elevated levels at the consumer’s taps if they were to be mobilized. The data indicate distinctly nonconservative behavior for many inorganic contaminants in drinking-water distribution systems. This finding suggests an imminent need for further research into the transport and fate of contaminants throughout drinking-water distribution system pipes, as well as a re-evaluation of monitoring protocols in order to more accurately determine the scope and levels of potential consumer exposure.

Introduction Background. Scales form on drinking-water piping through a combination of in situ or upstream corrosion processes, post-treatment deposition of solid materials resulting from treatment, and deposits formed by reaction of the finished water and piping with natural substances passing through the treatment plant. The nature and composition of scales in a drinking-water distribution system are a function of both the water quality and the metal pipe underneath. A voluminous body of research has documented the importance of corrosion control for regulated contaminants, such as lead * Corresponding author phone: 513-569-7412; fax: 513-5697172; e-mail: [email protected]. † United States Environmental Protection Agency. ‡ Pegasus Technical Services, Inc. 10.1021/es702488v CCC: $40.75

Published on Web 01/23/2008

 2008 American Chemical Society

(Pb) and copper (Cu),which relies upon the manipulation of water chemistry to develop passivating or immobilizing solid phases on the pipe surfaces. Research has shown that water constituents other than Pb and Cu, such as aluminum (Al), radium (Ra), and arsenic (As), also accumulate in scales on distribution system pipes (1–4). Recently, many prior studies have been reviewed and summarized (5). In some cases, the adsorption and accumulation of constituents, such as Al, could serve to help form corrosion-inhibiting passivating films or diffusion barriers (6–10). Recently, however, concern has begun surrounding the stability of scales and their potential for unpredictable, acute elevations of contaminants in distribution system locations not currently included in regulatory monitoring programs. Under normal circumstances, trace amounts of known contaminants do not pose a health risk; however, the release of accumulated contaminants in concentrated amounts could result in high levels at consumers’ taps (5, 11, 12). Release episodes of both regulated and unregulated contaminants from pipe scales into drinking water, including Pb, As, and Rn (from Ra), have been documented (4, 5, 11–15). The recent occurrences of contaminant releases in domestic water systems challenges the traditional assumption that inorganic contaminants behave conservatively in a distribution system—that is, that inorganic contaminants hold the same concentration from the end of treatment to the tap (3, 5, 16). Current U.S. drinking-water regulations for inorganic contaminants focus on monitoring effective removal of source water contamination. In the United States, the National Primary Drinking Water Standards (NPDWS) protect public health by limiting the levels of contaminants in drinking water, and National Secondary Drinking Water Standards (NSDWS) regulate contaminants that may cause aesthetic or cosmetic effects in drinking water. With the exception of lead and copper corrosion control and asbestos fiber contamination, these regulations do not require sampling procedures that would be effective in identifying risks associated with the accumulation and re-release of inorganic contaminants or radionuclides from distribution system mains, service lines, or interior premise or building plumbing (17, 18). Scope of Study. This study was derived from our current research into the development of equilibrium chemical models to predict treatment impacts on lead corrosion. The limited objective of this present work was to use this valuable and unusual existing body of “samples of opportunity” to assess the likely co-occurrence of a broad suite of inorganic contaminants in lead pipe scales. Analytical approaches available included inductively coupled plasma-atomic emission spectroscopy (ICP-AES), inductively coupled plasmamass spectrometry (ICP-MS), continuous flow-cold vaporatomic absorption spectrometry (CV-AAS), and total carbon and total sulfur by combustion. The nature of the origin and distribution of these pipe samples precludes their use as either an objective quantitative estimate of national occurrence of metal accumulation either in lead pipes, or in pipe scales generally. These data, however, were intended to substantiate or contradict prior predictions of the need for concern about potential contaminant accumulation in the distribution system, in water systems meeting the Safe Drinking Water Act (SDWA) regulations. Experiments to estimate the mobility of scale contaminants in their respective water supplies were beyond the scope of this investigation. Sample Collection. Scale samples were collected from 91 pipe specimens made available by 26 different municipal water distribution systems across 8 states over a span of 16 VOL. 42, NO. 12, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4285

FIGURE 1. Geographic origins of lead pipe specimens from which samples were harvested. For the locations of the pipe specimens, systems distributing treated groundwater are plotted with a green dot, systems distributing treated surface water are shown with a red dot, and systems distributing a mixture of ground and surface water at some or all of the sample locations are shown with a blue dot. years, though most have been received within the past 5 years. Of the 26 different water distribution systems, samples were mainly exposed to treated surface waters for 15 of the systems, treated groundwater for 9 of the systems, and to a combination of treated surface water and groundwater for 2 of the systems. Figure 1 is a map showing the originating locations of the pipe specimens. The pipe samples used for this study were obtained as part of lead corrosion or simultaneous regulatory compliance studies. They do not represent a statistically valid random sample of the national population of lead service line scales. Notably, with the exception of some exceedences of the lead or copper action level, none of the drinking-water utilities represented by the samples were in violation of any other inorganic contaminant regulations.

Materials and Methods Sample Preparation. Upon receipt in the laboratory, prior to any processing, the ends of each sample pipe were plugged with rubber stoppers and all loose material was washed from the outer surface. The pipes were cut longitudinally to expose the inner wall using a band-saw with a fine-toothed carbonsteel blade. During this process, small lead particles often scattered throughout the pipe and became embedded in the existing scale. To the extent possible, these particles were blown off with a laboratory air jet or manually removed from harvested scale using an assortment of fine-tipped tools, such as forceps or dental picks. Some scale material was lost from some pipes during shipment and cutting, though any relatively loose scale material dislodged during transport was collected when the pipe ends were uncapped. Following photography, stereomicroscopic observations, and mineralogical descriptions, scale material was then harvested from the pipe in operationally defined layers. These different 4286

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 12, 2008

“layers” of scale were harvested using fresh or detergentcleaned and deionized water-rinsed soft brushes, various metal spatulas, and fine-tipped tools. The discrimination of a “layer” was mainly based on color, textural change, or both, in scale composition, as is illustrated by the photographs in Figure 2. The layers were sequentially numbered from outermost (L1) to innermost, with most pipes appearing to have 3 layers based on these criteria, with exceptions showing either fewer or more layers. The total scale thickness, as well as individual layer thickness, varied widely across samples, and sometimes also within a given pipe sample. Generally, total scale thickness varied from approximately 0.1 to several mm, suggesting strong dependence on water chemistry history. The scale volume harvested from each layer and in total was a function of the length of the pipe specimen and the mineralogical characteristics of the scale. Frequently, less than 20–50 mg of scale from a given layer was available for all analyses, though as much as 1–2 g was occasionally available for long pipe sections with thick scales. Scales harvested were manually ground with an agate mortar and pestle until able to pass through a 200-mesh (e75 µm grain size) stainless steel sieve. Aliquots of this material were utilized for both elemental analysis and X-ray diffraction analysis, while only the former is the subject of this study. Metallic Elemental Scale Analysis. All scale analyses for which results are reported here were conducted by the U.S. Geological Survey (USGS). Because the volume of scale material was often limited, the application of analytical techniques were prioritized until the scale aliquot was consumed. The highest priority was for analysis for forty elements via ICP-AES, using a Perkin-Elmer Optima 3000 simultaneous spectrometer. The ICP-AES analyses were performed after decomposition using a sequence of steps

FIGURE 2. Examples of layer structure in two different lead pipe scales: (a) a two-layer primarily PbO2 scale, with L1 a nearly continuous layer overlying L2; (b) a complex multilayered scale, with dissimilar outer material (L1) rich in Mn, Fe, and Al oxyhydroxides on top of primarily lead hydroxycarbonate layers 2 and 3 (at metal pipe interface). Scale ticks represent 1 mm. with hydrochloric, nitric, perchloric, and hydrofluoric acids at low temperature (19). The second priority analysis used ICP-MS conducted on a Perkin-Elmer Elan 6000 ICP-MS system. This method applied to rare earth elements and silicon, with solubilization of the sample material by sintering with sodium peroxide, leaching with water, and acidifying with nitric acid (20). The third priority analysis was total carbon and total sulfur by combustion. All total sulfur and total carbon analyses reported were performed at USEPA using a LECO model CS230 combustion furnace instrument (21, 22). Finally, when the scale volume permitted, mercury (Hg) was determined by CV-AAS using a Perkin-Elmer 3030B spectrophotometer. For mercury determinations, samples were digested with nitric acid and sodium dichromate (23). Table 1 presents the summary of elements of interest to this study, and their applicable U.S. drinking-water regulatory limits. No specific standard reference materials exist for water pipe corrosion deposits. Aside from having a very different amount of lead, NIST 2782 (Industrial Sludge) was deemed

to be reasonably suitable to represent many of the elements present in the scales and most potential analytical matrix interferences. Table 2 presents recovery data for the elements of interest during the course of the study. The low bias for chromium reflects the resistant form of the Cr in the reference material, and results on actual pipe samples are expected to be much more accurate. Due to lack of analytical sensitivity and small amounts of available material, consistently low reporting limits for the elements could not be achieved for all samples. Powder X-ray Diffraction (XRD). As part of other studies, all samples included in this study had previously been analyzed by powder X-ray diffractometry. A Scintag SDS 2000 theta-theta diffractometer was used for almost all of the samples. This unit is equipped with a Peltier detector and a Cu KR X-ray tube, and the scans were performed in stepscan mode. Operating conditions were generally 35–40 kV, 40 mA, 0.02° 2θ step size, and a 1-s hold time. The remaining samples were analyzed using a PANalytical ’Pert with CoKR VOL. 42, NO. 12, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4287

TABLE 1. Drinking-Water Contaminant Elements and Their Respective Regulatory Status (Bismuth, Nickel, Sulfur, and Tin Are Also Contaminants Considered in This Study but Are Not Regulated or Present on the CCL) contaminant list

contaminant

symbol

NPDWSa As Ba Cd Cr

standard MCLb (mg/L) 0.01 2 0.005 0.1

arsenic barium cadmium chromium (total) copper lead mercury

Cu Pb Hg

aluminum copper iron manganese zinc

Al Cu Fe Mn Zn

1.3 (TTc action level) 0.015 (TT action level) 0.002 secondary standard (mg/L) 0.05-0.2 1.0 0.3 0.05 5

vanadium

V

N/A

NSDWSd

CCLe a

National Primary Drinking Water Standards. b Maximum Contaminant Level. c Lead and copper are regulated by a treatment technique that requires systems to control the corrosiveness of their water. If more than 10% of tap water samples exceed the action level, water systems must take additional steps. d National Secondary Drinking Water Standards. e Contaminant Candidate List.

radiation, PW3050/60 theta-theta goniometer, and PW3011/ 20 proportional detector. Data Analysis. The elemental data generated by the different methods for all layers of all samples with sufficient material were tabulated, validated, and stored in a Microsoft Access relational database. The outermost layer of scale in contact with the water (Layer 1) was considered to be presumably the most reactive, and those data are used preferentially in this evaluation. If there was insufficient material in Layer 1 for the chemical analysis, Layer 2 data were used. The subset of data for the layers of focus was analyzed using frequency distribution plots created by the frequency plot transform available in Sigma Plot software, and summary statistics were calculated in Microsoft Excel. In the interest of determining general contaminant levels, ranges, and trends, statistical analysis of the data focused on summary distribution statistics and mean values for the elements of interest to this study from those samples that had analysis results at or above the reporting limit for that element. The reporting limits in this study were determined by elemental analysis results that showed finite values reported by the laboratory, while the detection limits were defined here by elemental analysis results that included the concentration of the target contaminant at concentrations below the reporting limit. Detection limits and reporting limits varied by element, as expected, and occasionally by run, as a function of digestion, scale volume, and dilution effects. The reasons for employing this method of data analysis were various. Primarily, however, the ability to conduct analyses is dependent upon an adequate amount of scale available for harvest from a given layer, and the need to do comparisons of scale deposit layers that were as equivalent as possible across sites. The concentrations of contaminants reported here are indicative of the contaminant concentrations presumed to have most strongly interacted with drinking water flowing in the pipes. The data analysis was restricted to a subset of elements that were of particular interest based 4288

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 12, 2008

on either their regulation as drinking water contaminants, or as probable important species playing a major role in corrosion control interactions.

Results Contaminant Concentrations. Contaminants are separated into four arbitrary categories reflecting approximate orderof-magnitude groupings, based on average concentrations. Average concentrations ranging from 0 to 99 mg/kg are categorized as “minimal”, from 100 to 999 mg/kg are categorized as “minor”, from 1000 to 9999 mg/kg are categorized as “moderate”, and g10 000 mg/kg are categorized as “major.” Table 3 shows a summary of the categories of contaminants and some of their respective descriptive distribution statistics. Frequency distributions are shown in Figures 3and 5. Complete results from elemental analyses of the 91 scale samples can be found in the Supporting Information. With the exception of three samples, Pb is the largest component in all the samples analyzed. This was expected considering all scale samples were harvested from Pb pipes. The exceptions are a sample with Al as its largest component, a sample with Fe as its largest component, and a sample with Mn as its largest component. Minimal Contaminants Group. The concentration of As, Cd, Cr was over the reporting limit for 37, 34, and 81 of the 91 samples, respectively. Forty-eight, 53, and 9 results for As, Cd, and Cr, respectively, were above the detection limit but below the reporting limit. All samples contained Cr levels above the detection limit. Figure 2 shows the occurrence frequency across the concentration range (mg/kg) for results of As, Cd, and Cr that are above the reporting limit. Samples contain As concentrations as high as 426 mg/kg, Cd concentrations as high as 308 mg/kg, and Cr concentrations as high as 1290 mg/kg. Concentrations above the reporting limit for Hg exist in 23 of 29 samples, with concentrations up to 2 mg/kg. Minor Contaminants Group. The contaminants Ba, Bi, Ni, and U comprise the minor contaminants group, with ICP analysis results for Ba, Bi, and Ni showing results over the reporting limit for 87, 49, and 65 of the 91 samples, respectively. Three, 28, 24, and 52 results for Ba, Bi, Ni, and U, respectively, are above the detection limit but below the reporting limit. All 91 samples contained Ba concentrations above the detection limit. Results for Bi were not available for 13 samples, 1 sample had no results for Ni, and 38 of the 91 samples had no results for U. Figure 3 shows the occurrence frequency across the concentration range (mg/ kg) for concentrations above the reporting limit. Ba concentrations as high as 2850 mg/kg were found in the pipe samples, while Bi concentrations were as high as 1960 mg/ kg, and Ni concentrations were as high as 1800 mg/kg. Moderate Contaminants Group. ICP analysis results for Cu, Sn, V, and Zn show results over the reporting limit for 91, 49, 78, and 85 samples, respectively. Sulfur analysis was not conducted on all samples used in this study, as a result of limited sample volume. Nineteen results for Sn, 10 results for V, and four results for Zn are above the detection limits but below the reporting limits. All samples contained Cu concentrations above the reporting limit. Tin could not be analyzed in 20 samples. Cu concentrations were observed to be as high as 42 600 mg/kg, S concentrations were as high as 5972 mg/kg, Sn concentrations were as high as 9440 mg/ kg, V concentrations were as high as 22 000 mg/kg, and Zn concentrations were as high as 24 700 mg/kg. Other research in this laboratory using X-ray absorption near-edge spectroscopy (to be reported in detail elsewhere) revealed that a primary mechanism of vanadium accumulation was in the form of the mineral vanadinite (Pb5(VO4)3Cl). Major Contaminants Group. The contaminants Al, Fe, Mn, and Pb, which compose the major contaminants

TABLE 2. Measurement Summary for NIST2872 Used for Analytical Quality Assurance element Al Fe Mn Pb Cu Zn V Ba Ni As Cd Cr

true value

95% confidence value

mean observed value

observed std. dev.

weight % 1.37 26.9 0.03 0.0574 0.2594 0.1254 mg/kg 80 254 154.1 166 4.17 109

weight % (0.09 (0.7 NA (0.0011 (0.0052 (0.0196 mg/kg (10 (24 (3.1 (20 (0.09 (6

weight % 1.4 26.8 0.036 0.064 0.26 0.132 mg/kg 81 365 153 167 4.5 79

weight % 0.058 0.346 0.009 0.005 0.009 0.005 mg/kg 3.1 214 5.77 5.77 2.2 9.5

TABLE 3. Contaminant Categorization and Some Summary Statistics contaminant category minimal (0 to 99 mg/kg)

minor (100 to 999 mg/kg)

moderate (1000–9999 mg/kg)

major (g10 000 mg/kg)

contaminant As Cd Cr Hg Ba Bi Ni U Cu S Sn Zn V Al Fe Mn Pb

minimum (mg/kg) 10.6 2 3 0.03 1 0.6 3 19 35 5 10 12 29 0.6 7 25600

group, were present in the scale samples. ICP analysis results for Al, Fe, Mn, and Pb show results over the reporting limit for 72, 89, 91, and 91 samples, respectively. Four results for Al are above the detection limit but below the reporting limit. All samples contained Mn and Pb concentrations above the reporting limit. Samples contain Al concentrations as high as 44 000 mg/kg, Fe concentrations as high as 578 000 mg/kg, Mn concentrations as high as 177 200 mg/kg, and Pb concentrations as high as 915 000 mg/kg.

FIGURE 3. Frequency distribution of concentrations for minmal and minor contaminants.

maximum (mg/kg)

median (mg/kg)

average (mg/kg)

426 49 87 308 6.4 28 1290 40 83 2 0.18 0.33 2850 105 199 1960 320 422 1800 57 129 reporting limit